Computer Science · Hardware and Architecture
Limitations in predicting defect level based on stuck-at fault coverage
Jihun Park, M. Naivar, R. Kapur, M.R. Mercer, T.W. Williams
See on the mapAbstract
The stuck-at fault model has been used over decades as a guide to the test generation process and as an evaluation mechanism for the quality of the test set. As demands on quality have increased, the use of the stuck-at fault model as a predictor of the defect level has been questioned. This paper provides some insight on the issue and shows the limitations of using the stuck-at fault coverage to predict the defect level. The authors demonstrate that as defect level decreases the uncertainty of the estimate grows.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">></ETX>